Cancer Somatic Mutations And Clonal Evolutionary Dynamics

High error rate is a key issue to detect somatic mutations. Firstly, we used two monomorphic natural populations to check the correlation of error pattern between ABI-SOLiD and Illumina-GA. We found dual application reduced the false positive rate sharply. Then we developed a method to estimate polymorphism (θ) from pool-line data by combining two platforms. The results showed the estimation from dual application is more accurate than single platform in the same sequencing depth. Secondly, we developed a feature-selection method (CASpoint) based on machine learning technique to detect somatic SNV. The results showed CASpoint outperformed other6somatic SNV detectors in six different datasets from three cancer genomic/exome projects.For studying the clonal evolutionary dynamics in intrahepatic metastatic Hepatocellular Carcinomas (HCCs), whole-exome sequencing was applied to12such cases with84sections sequenced and/or genotyped. In two of them, the tumors are of independent origins and migration plays no role. For the rest, the clonal relationships of multifocal tumors often follow a spatial pattern in which cell migration precedes clonal diversification&growth. The percentage (75%) of early emergence of intrahepatic metastasis, timed by mutation accumulations, is unexpectedly high. The interpretation is corroborated by frequent mutations in genes of the extracellular matrix components (e.g., C017A1and LAMAl) and cell adhesion molecules (PTPRT and PCDH15); both classes pertain to cell movement. By proposing a new method for inferring adaptive evolution in cell lineages, we showed that invasive migration characterizes7of the8informative HCC cases. With invasion, different tumors, or different sections of the same tumor, may differ in their underlying causes of growth. Each proliferation event is driven by a different set of mutations, many of which are not common. Since migration by itself rarely confers an advantage to cells, its selective advantage is derived from a mutual reinforcement of tumor-growth and cell-motility mutations that creates positive feedbacks in fitness. In multifocal tumors, this mutual reinforcement leads to a process of adaptive diversification that accelerates as tumors evolve. The clonal evolutionary dynamics is both evolutionarily interesting and clinically relevant.We performed exhaustive microdissection for a slice of a primary HCC and obtained288 sections with average sample size of～33000cells. We detected332nonsynonymous/splice-site SNVs from whole-exome sequencing data of15tumor sections and a normal section. Further verification and genotyping of33sections characterized the spatio-temporal evolutionary dynamics during tumor growth.